What is fine-tuning?
Model fine-tuning is further training of an existing AI model on a smaller set of your own examples, so it handles a specific task or style more reliably. It changes how the model behaves, not which current facts it knows.
Also called: fine tuning, model fine-tuning
Why it matters
Fine-tuning costs time and data: you need many good examples, and you repeat the work when the base model changes. For most company use cases, better prompts or RAG get results sooner. It pays off for narrow, high-volume tasks where a smaller tuned model can replace a larger, slower one.
1What you needA set of example inputs with ideal outputs, cleaned and checked by someone who knows the task.
2Fine-tuning vs RAGUse RAG when answers depend on documents that change; use fine-tuning for a fixed format, tone or classification.
3In machine learningThe same idea applies to image and speech models: start from a trained model instead of training from zero.
ExampleA logistics company in Almaty sorts a high volume of customs and transport documents into a few dozen types. A large model does it well but slowly. After fine-tuning a smaller model on labelled past documents, the sorting runs faster and costs less per document, and the team checks a sample every week.
On this topic
What we do
Related terms
LLMA large language model: an AI model trained on huge amounts of text to read and write language.RAGA way to make an AI model answer from your own documents by fetching the relevant passages first.Prompt engineeringWriting and testing the instructions given to an AI model so it produces reliable, usable output.
Need to put this in place?
Describe the task, we will name the scope, price and timeline. The price is fixed in the contract.
Your selectionAI agent developmentfrom 480 000 ₸
The team gets your request; we reply during business hours.Price fixed in the contract before you pay.No calls unless you ask.